Bacteremia at a Danish university hospital during a twenty-five-year period (1968-1992).

In the 25-year period 1968-92, 3,317 out of 477,420 patients admitted to Frederiksberg Hospital experienced 3,491 episodes of bacteremia. Enterobacteriaceae dominated as causative agents (57%), following by Gram-positive cocci (31%) and anaerobes (7%). Polymicrobial bacteremia was found in 8% of the episodes. The incidence of Enterobacteriaceae bacteremia culminated in the middle (1978-82) of the period (4.7/1,000 admissions) and decreased during the last decade. Gram-positive bacteremia increased throughout the period (from 1.8 to 2.9; p < 0.001), due mainly to increasing incidences of bacteremia caused by non-hemolytic streptococci, Streptococcus pneumoniae and coagulase-negative staphylococci. Bacteroides fragilis accounted for a rising incidence of anaerobic bacteremia (from 0.3 to 0.7; p < 0.05). Clinical data were available for the 2,599 bacteremic episodes in the 20-year period 1968-87. 59% of these were hospital acquired. Of those, 38% were associated with indwelling catheters, mainly bladder catheters (28%) and i.v. lines (7%). The urinary tract dominated as source of bacteremia (46%), followed by the respiratory (11%) and the gastrointestinal tract (9%). Half of the patients had predisposing underlying diseases, most frequently malignancies (20%) and diabetes mellitus (7%). The mortality rate related to bacteremia decreased from 25% to 11% (p < 0.001). More than half (55%) of the fatal cases related to bacteremia occurred within the first 2 days after the first positive blood culture was obtained. Logistic regression analysis defined 7 variables that independently influenced the outcome related to bacteremia: age, source, culture verification of source, shock, body temperature, leukocyte count and empiric antibiotic treatment.(ABSTRACT TRUNCATED AT 250 WORDS)